标签：#AI

AI design system output is approximate by default. Wrong border radii, raw hex values, inconsistent tokens across 60 components. The fix isn't better prompts. Here's the structural change that made it exact using Figma's REST API. The fourth time I manually corrected the same border radius mistake in an AI-generated component, I stopped and asked why this kept happening. Not "what prompt would fix this?" The deeper question: why does every AI tool I tried get the structure right and the values wrong? The button was correct. The variants were there. The layout matched the Figma spec. But borderRadius: 8 when it should be borderRadius: '8px' . A spacing gap of 8 when the spec said 6 . The color #3B82F6 sitting in the file where semantic.button.primary should be. None of it wrong in a way that breaks the build. All of it wrong in a way that breaks the design system. After hitting this wall enough times, I realized the problem wasn't the AI. It was the question I was asking it. Why AI keeps generating the wrong Figma design tokens When you give an AI tool a Figma screenshot and ask it to produce a component, it does something reasonable: it interprets what it sees. The structure, the layout, the hierarchy - it gets most of that right. What it cannot get right is the token mapping. The AI doesn't know your semantic token file. It doesn't know that #3B82F6 maps to semantic.button.primary in your codebase. It doesn't know that your MUI setup multiplies numeric border radii by 4, which means borderRadius: 8 renders at 32px instead of 8px . So it approximates. Here's what that looks like in practice: What AI produces What the spec requires Why it's wrong borderRadius: 8 borderRadius: '8px' MUI multiplies numeric values by 4 gap: 8 gap: 6 Spacing value not extracted from Figma color: '#3B82F6' semantic.button.primary Raw hex instead of semantic token fontSize: 14 variant="MD_Medium" Typography token not resolved Across one component, these deviations are small. Across 60 comp

We wired our LLM eval suite into Datadog over about four months. Most of the panels we built got deleted. These are the five that stayed, and the metrics that feed them. TL;DR: We run an LLM-as-judge eval suite on every PR that touches a prompt, and we ship the results to Datadog as custom metrics. The dashboard started with fourteen panels. We kept five. The one that catches the most real regressions is per-criterion pass-rate split out by judge criterion, not the single rolled-up pass-rate number, because an aggregate of 91 percent hid the fact that one criterion had dropped from 0.95 to 0.62. Below are the metrics we emit, the Python that submits them, the monitor config we alert on, and the panels we tried and dropped. Some context on the setup so the rest makes sense. We are a Series-C dev-tool startup. We have a handful of prompts in production that do real work (classification, extraction, a summarization step in an agent loop). Each one has an eval set of tagged examples, somewhere between 80 and 400 per prompt. The judge is a separate model call that scores each output against a rubric. We run the suite in GitHub Actions. The eval job emits metrics to Datadog at the end of every run. Backend service health was already in Datadog, so putting eval data next to it meant one place to look during an incident instead of two. 1. Emit per-criterion pass-rate, not just the rolled-up number This is the one that earns its place. Our judge scores each output against multiple criteria. For the extraction prompt it is four: correct fields, no hallucinated fields, format valid, no refusal. Early on we only emitted one number, prompt_eval.pass_rate, the fraction of examples that passed every criterion. That number is fine for a smoke test and useless for debugging. The problem showed up on a prompt change that looked clean. Overall pass-rate went from 0.93 to 0.91. Two points. Nobody would block a PR on two points. But underneath, the "no hallucinated fields" criterion had

80%. That is the share of code currently being merged into Anthropic's production systems that was written by Claude. Not code-reviewed. Not pair-programmed. Written. In February 2025, when Claude Code launched, that number was in the low single digits. Sixteen months later, the company decided that data point — and the trajectory behind it — was worth a public warning. On June 4, 2026, Anthropic published "When AI Builds Itself," a research paper co-authored by Marina Favaro, head of the Anthropic Institute, and Jack Clark, one of the company's co-founders. It was the first major publication from the Anthropic Institute since its founding in March 2026. The paper did two things simultaneously: disclosed internal productivity data that most AI companies keep private, and called for a global mechanism to slow or pause frontier AI development before the process becomes self-sustaining without meaningful human direction. The data came first. The policy recommendation followed from it. Here is what the numbers actually show and why every developer building on AI infrastructure today should read this carefully. The Productivity Curve Nobody Predicted Anthropic published a chart of engineering output per engineer, indexed to a baseline from 2021–2024. The curve is flat for four years. Then Claude Code shipped in February 2025. The multiplier progression from that point: 1.2x, 1.5x, 1.9x, 2.5x. By Q1 2026: 5.8x. By Q2 2026: 8x. The typical Anthropic engineer is now merging eight times as much code per day as they were in 2024. Not 8% more. Eight times more. That is not a productivity improvement — it is a different category of output from the same headcount. To understand what drives the number, you need to understand what Claude Code actually does inside Anthropic's engineering workflows. The tool was built for and by engineers working on frontier AI systems — which means the tasks it handles are not boilerplate CRUD endpoints. Claude is writing test harnesses for novel m

Cycle tracker will now notify women when their cycle patterns are suggestive of perimenopause.

The idea behind the new "Siri AI" is to turn the assistant from a voice controlled assistant into an AI companion that can do a lot more.

Amazon is expanding its print-on-demand features to AI-generated designs created using Alexa for Shopping for products like T-shirts, water bottles, and hoodies. Shoppers can use text prompts to generate images that are then printed on to blanks for sale on Amazon. They can then share the link to the design so other people can buy […]

Two years after first revealing its plans for Apple Intelligence and a smarter Siri that never fully materialized, at WWDC, Apple just revealed a new set of AI features and a smarter, more personalized Siri. Apple calls Siri AI an "entirely new version of Siri" and says it's both more conversational and more capable than […]

If you’re a user—owner?—of this cryptocurrency, this is important: On May 29, the security researcher Taylor Hornby found a critical vulnerability in Zcash Orchard privacy pool using Claude Opus 4.8. The Zcash team hired Hornby specifically to look for this kind of issue. He found one fast enough to be embarrassing. The Orchard pool is the newest and most advanced shielded transaction system in the cryptocurrency Zcash. Introduced in 2022, it allows users to send and receive ZEC while keeping transaction details private. It uses zero-knowledge proofs to validate transactions without revealing amounts or participants. The bug: a specific check that was supposed to validate transaction inputs wasn’t actually enforcing the rules it appeared to enforce. An attacker could have exploited the flaw to feed false inputs into that check and generate ZEC from nothing, with the zero-knowledge proof system blessing the fraudulent transaction as valid...

Google is rolling out "across the board" updates to NotebookLM. The AI-powered note-taking app now uses Google's upgraded Gemini 3.5 model, which will allow it to respond with "more accurate and reliable information," according to a blog post on Monday. Launched in 2023, NotebookLM allows you to interact with your notes and sources using AI, […]

Per-request attribution starts with five fields on every call: provider, model, input tokens, output tokens, and ownership tags such as team, feature, and customer. A monthly vendor bill cannot explain why one feature, one tenant, or one prompt template suddenly became expensive. Request-level math can. As of June 8, 2026, OpenAI lists GPT-5.4 mini at $0.75 per 1M input tokens and $4.50 per 1M output tokens, while Anthropic lists Claude Sonnet 4 at $3 and $15 respectively. Gateway logs are useful, but they rarely solve AI cost tracking per feature unless you enrich them with business context and retry metadata. The practical operating model is simple: calculate cost on every request, attach ownership dimensions, then roll the data up into team, feature, and customer views. If you are searching for "LLM cost attribution per request," you are usually already past the basic billing problem. You can see your OpenAI or Anthropic invoice, but you cannot answer the questions finance and engineering actually care about: which feature drove the spike, which team owns it, which customers are unprofitable, and which prompt or model change caused the jump. That is why per-request attribution matters. It turns AI spend from a monthly surprise into an operational metric you can act on in the same day. Why LLM cost attribution per request matters now According to the FinOps Foundation's 2025 State of FinOps report, 63% of respondents now manage AI spending, up from 31% the year before. That jump is the real signal. AI cost is no longer a side bucket inside cloud spend. It is becoming a first-class FinOps workload. For teams spending $5,000 to $50,000 per month on LLM APIs, averages break down quickly. A support assistant, an internal coding copilot, and a customer-facing generation feature can all hit the same vendor account while having completely different margins, latency targets, and prompt shapes. If you only look at total spend by provider, you lose the unit economics. Per-r

OpenRouter is a unified API gateway to hundreds of language models from providers such as OpenAI, Anthropic, Google, and Meta. You use one API key and one billing surface, and swap models by changing a provider/model slug. OpenRouter exposes a Chat Completions -compatible HTTP API. This post shows three Node.js integration paths: the official @openrouter/sdk , the openai package with baseURL , and the Vercel AI SDK with @openrouter/ai-sdk-provider . For deeper patterns on each stack, see the Chat Completions API , OpenAI Responses API (OpenAI direct only), and Vercel AI SDK posts. Prerequisites OpenRouter account API key Credits or billing enabled as needed Node.js version 26 Install packages for the path you use: @openrouter/sdk ( npm i @openrouter/sdk ) openai ( npm i openai ) ai and @openrouter/ai-sdk-provider ( npm i ai @openrouter/ai-sdk-provider ) Configuration Read credentials from the environment in production. Variable Purpose OPENROUTER_API_KEY Bearer token from OpenRouter settings OPENROUTER_MODEL Default model slug, for example openai/gpt-5.5 OPENROUTER_SITE_URL Optional site URL sent as HTTP-Referer for rankings on openrouter.ai OPENROUTER_SITE_TITLE Optional app name sent as X-OpenRouter-Title Model IDs use the provider/model format, for example openai/gpt-5.5 , anthropic/claude-opus-4.8 , or google/gemini-3.1-flash-lite . Browse the full catalog at openrouter.ai/models . The examples below use openai/gpt-5.5 , matching the model in the other LLM posts in this series. Override it with OPENROUTER_MODEL when you want a different model. @openrouter/sdk OpenRouter's official TypeScript SDK is type-safe and generated from the OpenAPI spec. Client setup import { OpenRouter } from ' @openrouter/sdk ' ; const client = new OpenRouter ({ apiKey : process . env . OPENROUTER_API_KEY , httpReferer : process . env . OPENROUTER_SITE_URL , appTitle : process . env . OPENROUTER_SITE_TITLE , }); Basic integration const response = await client . chat . send ({ chatReques

A new feature in the Amazon Shopping app allows users to generate designs with Alexa, then print them on products like T-shirts, hoodies, and tumblers.

Apple's WWDC nears: Here's what you can look forward to.

In multi-turn agent loops, the full context re-sends on every API call. A tool result added at turn 3 gets billed again at turns 4, 5, 6, 7... forever. Most of it is never read again. Standard observability tools tell you the total token count. They never tell you what's in there or how much of it is waste . That's what ContextLens fixes. What it does ContextLens is a diagnostic profiler for LLM agent context windows. It: Decomposes the context window into regions: system prompt, tool schemas, tool results, retrieved chunks, user messages, assistant messages Tracks which blocks get re-billed across turns using SHA-256 content hashing Runs 5 waste detectors and ranks findings by dollar cost Prints a concrete one-line fix for each finding Renders an interactive D3 treemap report as a self-contained HTML file No API key required. Works offline on saved traces. The five detectors Detector What it finds Duplicate Same block re-sent verbatim across multiple turns Near-Duplicate >85% Jaccard similarity between distinct blocks Stale Tool Result Tool output never referenced by a later assistant message Unused Tool Schema Tool defined every turn but never called Redundant Retrieval Retrieved chunk with <15% overlap with model output ---Run the built-in demo (simulates a 30-turn agent loop, no API key needed): python -c "import contextlens; contextlens.demo()" python examples/demo.py Live capture — Anthropic import anthropic import contextlens as cl client = anthropic.Anthropic() with cl.capture_anthropic(client, model="claude-3-5-sonnet-20241022") as collector: for turn in range(20): client.messages.create( model="claude-3-5-sonnet-20241022", max_tokens=1024, system="You are a helpful assistant.", messages=build_messages(turn), ) report = cl.analyze_trace(collector.build_trace()) print(f"Recoverable waste: {report.recoverable_tokens:,} tokens (${report.recoverable_cost_usd:.4f})") Live capture — OpenAI import openai import contextlens as cl client = openai.OpenAI() with cl.ca

This weekend, AgentTrust ID went live in production. As of today, all five SDKs are published: pip install agenttrustid npm install @agenttrustid/sdkgo get github.com/agenttrustid/sdk/go cargo add agenttrustid # Maven / Gradle # id.agenttrust:agenttrustid:0.3.0 The SDKs are open source under Apache 2.0 at github.com/agenttrustid/sdk . The hosted platform is running at app.agenttrust.id in a controlled beta. Why I built this AI agents broke the assumptions that machine-to-machine security was built on. An API key answers one question: who is calling. It asks it once, at the door. An agent decides its next action at runtime, from context nobody wrote by hand. The same agent that summarized a document a second ago might now try to email it, delete it, or chain a task to another agent. A credential that only proves identity has no opinion about any of that. Agents need a decision at the action boundary : should this specific action happen, right now, on whose behalf . Answered at runtime, every time, with an audit trail and a kill switch. What's running Everything below is live in production today, not a roadmap: Per-action authorization. Every consequential action passes a pre-flight check. The Guardian pipeline routes each action by risk: deterministic rule checks for the common path, a policy engine for mutations, and AI-backed review for destructive operations. Fail-closed where it counts. Opaque, instantly revocable tokens. Credentials are at_ references with no standing authority of their own . The server decides on every use, so revocation is one call, effective immediately. Scoped delegation. When one agent hands work to another, the grant narrows instead of copying : subset scopes, independent TTLs, independently revocable, bounded chain depth. Read-only sessions with time-boxed elevation. Sessions start safe and rise only on approval, for a bounded window, then revert on their own. One model across surfaces. MCP tools, agent-to-agent calls, and direct API inte

https://www.youtube.com/watch?v=VD9zEKQEJxo 这视频深入拆解了人工智能强化学习之父、图灵奖得主理查德·萨顿（Richard Sutton） 在2026年5月共同发表的一篇仅有7页、零算法、零跑分的哲学立场论文。这篇论文提出了 “行动认知 AI”（Enactive Artificial Intelligence，简称 Enactive AI）的概念，并在科技界和资本圈引发了巨大震动（甚至让红杉、英伟达、谷歌联合下注了11亿美元成立新公司）。 视频从 核心概念、哲学脉络、理论内在矛盾、认知科学质疑 以及 产业界的三路对赌 五个维度，极其详细地复盘了视频的所有核心内容： 一、 什么是“行动认知 AI”（Enactive AI）？ 视频强调，全网很多地方都把 Enactive （行动认知/生成认知）和 Generative （生成式 AI，如 GPT、Sora）混淆了，但两者的底层逻辑恰恰相反 [ 00:50 ]： 生成式 AI（Generative AI）： 核心是 续写和预测 。通过已有画面或文本，被动地去预测下一帧、下一个词长什么样 [ 01:07 ]。 行动认知 AI（Enactive AI）： 核心是 在互动中现生成认知 。认知不是大脑被动接收信号并建立静态世界模型，而是“你动了手，世界才向你显现” [ 01:47 ]。 > 举例： 人去拿杯子，不是眼睛先拍下一张静态照片让大脑去死算距离、角度 [ 01:53 ]，而是手往前探的过程中，随着角度、光影的实时动态变化，杯子的形状和可抓取性才在动作里一点点“长出来” [ 01:59 ]。 感知和行动硬死在一起，无法拆分。 这套理论源自认知科学中的 自创生（Autopoiesis）与自主性（Autonomy） [ 02:21 ]。它认为智能体应该像生物一样自我维持、组织，由内在生存需求去塑造感知，而不是一个干等着外部指令输入输出的机器 [ 02:24 ]。 二、 萨顿为什么要发这篇哲学论文？ 萨顿并不是一时性起，这是他为了对抗当前“大模型路线”打出的最后一张哲学底牌： 2019年《苦涩的教训》： 主张人类手写规则干不过堆算力、让机器自己学的通用方法 [ 02:47 ]。 2024年《大世界假设》： 真实世界远比静态内部模型复杂，智能体必须在运行中实时学习 [ 02:59 ]。 2025年《经验时代》： 人类数据是有限的，AI 必须靠自己生成自己的经验长大的 [ 03:12 ]。 2025年9月： 直指整个 AI 行业走错路，大模型堆数据去超智是死路一条 [ 03:19 ]。 这篇论文补上了最后一把火： 之前的论证全是算力、数据和复杂度的“机械账” [ 03:25 ]。而这一次，他第一次把强化学习（RL） 和 认知科学（行动认知）接在了一起，从本体论层面证明： 大模型路走不通，认识世界这件事本身，就只能通过行动和互动的经验来发生 [ 03:39 ]。 为此，2026年初论文共作者创办了 Ineffable Intelligence 公司，号称要造出完全不需要人类数据、靠自己学习的 AI，直接拿到了红杉、英伟达、谷歌 11 亿美元的巨额融资（估值 51 亿美元） [ 03:55 ]。 三、 论文隐藏的两大致命致命逻辑“回旋镖” 视频话锋一转，指出萨顿借来的这套哲学地基里，埋着两根砸中他自己的“大柱子”： 柱子 1：砸中了萨顿的“奖励假设”（自相矛盾） [ 04:35 ] 强化学习的号称教条： 奖励假设（Reward Hypothesis），即所有目标、意图都可以写成“最大化外部给定的标量分数” [ 04:53 ]。David Silver 甚至喊出“奖励就够了” [ 05:13 ]。 行动认知哲学的教条： 自主性（Autonomy），即什么是好坏、成败，标准必须从智能体随时会散架的“物理组织和生存危机”中自发长出来，不能由外部权威操控 [ 05:27 ]。 裂缝： 标准强化学习的奖励函数（Reward Function）是人类设计者用代码硬塞进去的（他律） [ 05:55 ]；而生物判断好坏是为了顶住熵增、维持结构不崩（自主） [ 06:11 ]。论文里作者自己也承认：强化学习的评估标准依然由外部奖励定义 [ 06:38 ]。 内驱动机能救场吗？ 比如好奇心驱动或求知驱动。视频认为不能，因为诸如“优化预测误差”的总结优化目标，依然是人类在架构层死死规定好的，根本不是智能体出于生存忧关的自发需求。没有真正的生命威胁，就没真正的意义生成 [ 07:12 ]。 柱子 2：砸中了萨顿自己的《苦涩的教训》 [ 07:49 ] 萨顿当年痛骂：研究者总忍不住把人类以为的思考结构（比如语法树、手工特征检测器）硬塞进 AI 架构里，这长期必被碾压 [ 08:13

Here's where you can watch Christopher Nolan's 'The Odyssey' in jaw-dropping IMAX 70mm.

Today I’m talking with Mustafa Suleyman, the CEO of Microsoft AI. And I’m actually going to keep today’s intro short — I’m working from my wife’s family farm this week, as you’ll see in the video, but also this is a real burner of an episode. We covered everything from Mustafa’s approach to training new […]

Summer Game Fest may be over, but the flood of gaming news isn't. Nintendo just announced that it'll be holding its next Direct showcase event on June 9th at 10AM ET. The stream, which you can find here, will be followed by a Treehouse Live event, and it sounds like it'll be pretty lengthy. "The […]

Two years ago, when Microsoft first revealed that it was bringing four Xbox-exclusive games to the PS5 and Nintendo Switch, it made the announcement far more complicated than necessary. That's not likely to improve anytime soon. In fact, things now seem more confusing than ever as the company tries to appease both fans and the […]